SIMULARE

Seediscussions,stats,andauthorprofilesforthispublicationat:http://www.researchgate.net/publication/260927664

TheEffectofSprayDistanceandScanningStepontheCoatingThicknessUniformityinColdSprayProcess

ARTICLEinJOURNALOFTHERMALSPRAYTECHNOLOGY·OCTOBER2013

ImpactFactor:1.49·DOI:10.1007/s11666-013-0002-0

DOWNLOADS

41

VIEWS

75

5AUTHORS,INCLUDING:

SihaoDeng

UniversitédeTechnologiedeBelfort-Mont…

11PUBLICATIONS12CITATIONS

SEEPROFILE

GhislainMontavon

UniversitédeTechnologiedeBelfort-Mont…

79PUBLICATIONS690CITATIONS

SEEPROFILE

Availablefrom:SihaoDeng

Retrievedon:22June2015

The Effect of Spray Distance and ScanningStep on the Coating Thickness Uniformity

in Cold Spray ProcessZhenhua Cai, Sihao Deng, Hanlin Liao, Chunnian Zeng, and Ghislain Montavon

(Submitted September 7, 2012; in revised form July 11, 2013)

In the process of cold spray applications, robot kinematic parameters represent significant influences onthe coating quality. Those parameters include: spray distance, spray angle, gun relative velocity tosubstrate, scanning step, and cycle numbers. The combined effects which are caused by their interactionsdetermine the coating thickness. The increasing requirements of coating productivity lead to theobjectivity of analyzing the effect of robot kinematic parameters. So it becomes necessary to optimize therobot trajectory for spraying process in order to obtain a desired coating thickness. This study aims atinvestigating the relationship between the coating profile and the spray distance, scanning step, andintroducing the basic principle of a software toolkit named thermal spray toolkit (TST) developed in ourlaboratory to generate the optimized robot trajectories in spray processes including thermal spray andcold spray. Experiments have been carried out to check the reliability of the simulated coating profileand the calculated coating thickness by TST.

Keywords coating thickness, coating profile prediction, coldspray, robot trajectory, scanning step, spraydistance

1. Introduction

Thermal spray technology consists of a group of dif-ferent processes which can elaborate coatings withmetallic or nonmetallic materials (i.e., pure metals, alloys,ceramics, polymers, and composites). These materials aredeposited in a molten, semi-molten, or even solid state onthe substrate (Ref 1). Cold spraying is an emerging coatingprocess. In this process, particles are injected into a highspeed gas jet and accelerated to a high velocity (300-1200 m/s) (Ref 2). The coating consists of particles withintensive plastic deformation during impact in a solid stateat a temperature well below the melting point of thismaterial (Ref 3).

As shown in literatures (Ref 4-6), the homogeneity ofcoating thickness is influenced not only by the singlecoating profile but also the robot kinematic parameters.The single coating profile is determined by the materialproperties, spray distance from the nozzle to the substrate,spray angle, substrate properties, the substrate deforma-tion caused by local heat transfer, etc. Series of publica-

tions have shown that single coating profile can becharacterized by mathematic formulae (Ref 6, 7, 9), whichoffer the possibility to simulate and optimize the effect ofrobotic kinematic parameters with mathematical software.If a large number of single coating profiles are summed up,the distribution of the summed profile can be approxi-mately considered as a continuous curve (Ref 6).

Some publications concerning with the similar subjectshave been considered in this article (Ref 4-10). Li et al.(Ref 11) analyzed the effect of spray angle on coatingcharacteristics using cold spraying, and reported that therelative coating efficiency was maximum at spray anglesranging from 80� to 90�; and when the spray angledecreased to approximately 40�, there was almost noparticle deposited on the substrate and the relative coatingefficiency tended to zero. Kout et al. (Ref 7) investigatedthe planning path-oriented spray-coating processes, whichrepresented an optimization method to compute andapproximate the desired coating thickness with relativecoating parameters. Fasching et al. (Ref 12) presented anapproach for spraying layers using robotic thermal spraysystem, they offered equations to optimize the spray angle,and thus to generate more accurate robot trajectories.

In cold spraying, the single scanned coating is normalrather narrow comparing with other thermal spray pro-cesses and the step between two scanning is normallysmall to guarantee a uniform coating thickness. But fewliteratures considered this problem. This study thusfocused on the effect of spray distance and scanning stepon the coating thickness uniformity in cold spray process

Based on the above mentioned conclusions, the singlecoating profile is simulated with symmetric Gaussian dis-tribution curve in this study, and combining the curve withthe optimized robot kinematic parameters offered by

Zhenhua Cai, Sihao Deng, Hanlin Liao and Ghislain Montavon,IRTES- LERMPS, Universite de Technologie de Belfort-Mont-beliard, 90010 Belfort Cedex, France; and Chunnian Zeng,Automation School, Wuhan University of Technology, WuhanChina. Contact e-mail: [email protected].

JTTEE5 23:354–362

DOI: 10.1007/s11666-013-0002-0

1059-9630/$19.00 � ASM International

354—Volume 23(3) February 2014 Journal of Thermal Spray Technology

Peer

Revie

wed

‘‘thermal spraying toolkit’’ (TST), a software packagedeveloped by LERMPS, the suitable coating thickness canbe obtained within the required tolerances. A conceptnamed flatness is added to illustrate the homogeneity ofcoated thickness, and the relevant simulated coatingthickness and flatness result which were calculated by TSThave been presented and displayed on the graphic inter-face of TST.

2. Experimental Conditionsand Operation Parameters

The kinematic parameters of cold spray process aredemonstrated in Fig. 1: spray distance is the distance fromthe nozzle exit of spray gun to the substrate surface;scanning step is the distance of two neighbor scanningpasses; spray angle is the angle between the centerline ofthe spray gun and the substrate surface; gun velocity is therelative velocity between spray gun and substrate.

The experiment was carried out on polished aluminumsubstrates. The aluminum 5056 (average particle size40 lm) feedstock powder is selected for this study. Anoptimized rectangular nozzle, designed by LERMPS,having an expansion ratio of 4.9 and a divergent length of170 mm was used. The nozzle has a throat section of2 9 2.7 mm and an exit section of 4.4 9 6 mm. Com-pressed air was used as the driving gas at a temperature of873 k. Argon was used as the powder carrier gas at apressure of 2.8 MPa. The spray gun is guided by an IRB2400 robot (ABB, Switzerland). In order to measureprecisely the single scanned coating profile, a coordinatemeasuring machine was used (Derby� ETALON accu-racy: 0.0001 mm). The robot program is simulated inRobotStudio�5.13 (Off-line programming software ofABB) before its application. During the spraying process,the spray distance is changed from 10 to 70 mm. The gunvelocity is kept to 10 mm/s in order to obtain a very thickcoating. Thus only one layer coating was applied on aplane substrate. The spray angle has been kept to 90�. Thepoints which are picked up by coordinate measuring ma-chine from the single coating profile are fitted as curves

and mathematic equations were deduced from thosecurves using Matlab to represent the coating profile inTST, and then the simulated coating profiles are comparedwith the experimental results.

3. Experimental Results and Analysis

The coating profile was measured after being sprayedby the cold spray system, different spray distances (10, 30,50, and 70 mm) are applied during the experimental pro-cess. The single coating profile is measured five times; theaverage single coating thickness is calculated as the meanvalue of those five measured coating thickness. In order tomathematically analyze the coating profile, the softwaretool Matlab for numerical computing is employed.According to the first visual analysis of the curve, theGaussian distribution was then chosen for further simu-lation because the symmetric characteristics of the curve.The simulation process can be described as follows: about30 points picked up from each coating profiles areimported to Matlab; the distribution of 2D coating profileis then approximately fitted as a Gaussian curve with aconstant coefficient. Relative coating profiles under dif-ferent spray distances are shown in Fig. 2. It can beobserved that the thickest profile appears at a spray dis-tance equal to 50 mm (nearly 1.5 mm).

Suppose r represents the standard deviation ofGaussian equation. l represents the mean value ofGaussian equation; Z(x) represents the value of Gaussianequation and also stands for the height of 2D coatingprofile; S(x) represents the surface of 2D coating profile; Kis the constant coefficient of Gaussian equation; l repre-sents the length of 2D coating profile.

According to the definition of Gaussian equation, 2Dcoating profile is represented by the following equation:

Fig. 1 Parameters of cold spray processFig. 2 Fitted 2D coating profiles under different spray distancesin Matlab

Journal of Thermal Spray Technology Volume 23(3) February 2014—355

Peer

Revie

wed

Z xð Þ ¼ K

rffiffiffiffiffiffi

2pp exp �ðx� lÞ2

2r2

" #

: ðEq 1Þ

From Matlab, it can be observed that the simulated curveis smoother and more symmetrical than the real coatingprofile, and is also easy to precisely analyze in terms ofmathematic parameters. The R2 is 98.92%, which meansthe fitting result is nearly perfect. The Gaussian equation

used by Matlab as the simulation function is presented asfollows:

Z xð Þ ¼ a1 � exp½�ððx� b1Þ=c1Þ2�: ðEq 2ÞWhen comparing with Eq 1, the following relationship isdeduced:

a1 ¼ Kffiffiffiffiffiffi

2pp

rb1 ¼ l c1 ¼

ffiffiffi

2p

r: ðEq 3Þ

Therefore the value l, r, K is calculated by the followingequations:

l ¼ b1 r ¼ C1ffiffiffi

2p K ¼ a1 �

ffiffiffiffiffiffi

2pp

r: ðEq 4Þ

According to the definition of Gaussian function, standarddeviation r represents the extent of Gaussian distribution.K influences the thickness of coating profile. With thecalculated values, the deposit height as a function thatrepresents the particle impact distribution on the alumi-num substrate under different spray distances can be ob-tained. Standard deviation values calculated underdifferent spray distances are represented in Fig. 3.

From Fig. 3, it can be observed that the spray distancehas a significant influence on the coating thickness: thevalues calculated in Fig. 3 verify that the thickest coatingprofile appears at a spray distance of 50 mm, which is1.325 mm.

In order to make further analysis, the deposition effi-ciency is considered. Under the same experimental con-ditions, the surface of each 2D coating profile canapproximately illustrate its deposition efficiency. Thesurface of 2D coating profile S(k) is represented by Eq 5:

Fig. 3 Standard deviation curve and maximum coating thick-ness

Table 1 Surface of 2D coating profiles under differentspray distances

Spray distance, mm 10 30 50 70

Surface of 2D profile, mm2 5.68 7.8 9.462 8.394

Fig. 4 Interface of thermal spray toolkit

356—Volume 23(3) February 2014 Journal of Thermal Spray Technology

Peer

Revie

wed

S xð Þ ¼Z

l

0

ZðxÞdx ¼Z

l

0

K

rffiffiffiffiffiffi

2pp exp �ðx� lÞ2

2r2

" #

dx

¼ K �Z

l

0

1

rffiffiffiffiffiffi

2pp exp �ðx� lÞ2

2r2

" #

dx: (Eq 5)

According to the definition of Gaussian function, thesurface under Gaussian curve is always equal to 1. Thenthe surface of 2D coating profile can be represented by:

S xð Þ ¼ K ðEq 6Þ

K is deduced by Eq 3. Thus, the surface of 2D coatingprofile of different spray distances can be calculated, asshown in Table 1.

From Table 1, it can be concluded that the maximumvalue of S(x) appears at a spray distance of 50 mm, whichmeans the best deposition efficiency can be obtained atthat spray distance; compared to 7.8 mm2, the S(x) valuedecreases a little at a spray distance equal to 30 or 70 mm(no more than 2 mm2); at a spray distance of 10 mm, thesurface of 2D coating profile decreases obviously, which is5.68 mm2; this means compared to the spray distance of50 mm, the coating surface decreases almost 40%.

From the above experimental results, the spray dis-tance is associated with relative particle distribution atimpact with the substrate, and therefore represents dif-ferent coating profiles on the substrate, so it has significantinfluence on the width and height of 2D coating profile.For the aluminum 5056 used in this study, there exists anoptimal spray distance which represents the best deposi-tion efficiency. If the spray distance is lower than thisoptimal value, the particles is not accelerated enough to

obtain the best deposition efficiency; furthermore, if thespray distance is above this optimal value, the particlevelocity is decreased sharply, which means the relativedeposition efficiency will be decreased.

By measuring and fitting the 2D coating profile, themathematic model is obtained, which offers possibility tosimulate the 2D coating profile in the software. Based onthe above work, a new function of TST is developed tosimulate the 2D coating profile and to calculate the finalcoating thickness.

4. Thermal Spray Toolkit

TST is a developed graphic toolkit which is based onRobotStudio� (product of ABB Company) to predictcoating thickness and relative flatness; it is developed in C#(developing environment of Microsoft� company) andwith MSChart [A chart that graphically displays data(Ref 13)]. The data exchange between TST and Robot-Studio� is based on API functions (Ref 14). The interfaceof this function is similar to an assistant system which isdesigned to provide operational parameters by reasoningabout knowledge library (Ref 15, 16). The interface is

Fig. 5 Coating profile

Fig. 6 Off-normal spray angle on half Gaussian curve

Journal of Thermal Spray Technology Volume 23(3) February 2014—357

Peer

Revie

wed

divided into two groups: parameter group, which permitsimporting the experimental parameters (e.g., materialparameters library) and required parameters (e.g., scan-ning step, spray distance), as well as exporting the calcu-lated parameters (e.g., coating thickness); graphic group,which permits displaying the 2D coating profile and thecalculated thickness curve, as shown in Fig. 4.

Under normal conditions (for example, constantmaterial, experimental parameters; some robotic param-eters), coating profile curve can be considered as a con-stant curve which is described by Eq 1. The coatingprocess is considered as the accumulation of some coatingfilms. For one layer coating, the final coating profile can beapproximated as a continuous curve (Fig. 5). Thereby the

Fig. 7 Coating thickness simulated under different scanning steps. (a) Scanning step = 2 (mm). (b) Scanning step = 3 (mm). (c) Scanningstep = 5 (mm). (d) Scanning step = 7 (mm)

358—Volume 23(3) February 2014 Journal of Thermal Spray Technology

Peer

Revie

wed

Fig. 7 continued

Table 2 Coating thickness calculated based on different scanning steps

Scanning step, mm Coating thickness max, mm Coating thickness min, mmAverage

thickness, mm Flatness, %

2 1.22 1.22 1.22 1003 0.81 0.81 0.81 1004 0.61 0.609 0.61 99.825 0.495 0.481 0.488 97.176 0.43 0.38 0.4 87.957 0.39 0.292 0.34 74.878 0.39 0.216 0.3 55.38

Journal of Thermal Spray Technology Volume 23(3) February 2014—359

Peer

Revie

wed

following equation is proposed to simulate the final coat-ing thickness:

F xð Þ ¼ m½Z1 xð Þ þ Z2 x� pð Þ þ . . .þ Zn x� n � pð Þ�

¼ mK

rffiffiffiffiffiffi

2pp exp � x� lð Þ2

2r2

" #(

þ K

rffiffiffiffiffiffi

2pp exp � x� p� lð Þ2

2r2

" #

þ . . .þ K

rffiffiffiffiffiffi

2pp exp � x� n � p� lð Þ2

2r2

" #)

:

ðEq 7Þ

where F(x) is the coating thickness value, x is the abscisevalue, p is the scanning step which represents the lineardistance of two neighbor coating peaks, n is the nthcoating profile, m is the cycle number.

As shown in Fig. 5, starting from the second pass, off-normal spray angles for subsequent passes will appear. Formathematically analyzing the value of off-normal sprayangles, half the Gaussian curve (standard deviation: 3) isdivided into small sections with interval length of 0.1 mmin horizontal axis. The first derivative of the function Z(x)is the slop of the tangent to the function at each points x,so the off-normal angle can be calculated by:

AngleðxÞ ¼ arctan z0

xð Þ� �

� 180=p�

�

�

�

�

�

¼ arctan � xffiffiffiffiffiffi

2pp

r3e�

x2

2r2

� �

� 180=p

�

�

�

�

�

�

�

�

: ðEq 8Þ

Figure 6 shows the maximum off-normal spray angle isless than 1.6� at a standard deviation of 3, which confirmsthat the simulation equation and the simulated result canbe accepted in certain tolerance.

The simulated results can be directly obtained from theinterface of TST (Fig. 7), which helps to determine therobot kinematic parameters before carrying out theexperiment. More accurate data are shown in Table 2;coating thickness is calculated based on different scanningsteps. Flatness represents the percentage of minimum

coating thickness and the maximum coating thickness,which stands for the homogeneity of coating surface. Inthis table scanning steps are changed from 2 to 8 mm tocalculate average thickness and relative flatness, and tofind the relationship between scanning step and averagecoating thickness.

From Table 2, it can be observed that the averagethickness decreases with the increment of scanning step.The calculated flatness of coated surface is 100% whenscanning steps are less than 3 mm; this means the calcu-lated coated surface is absolutely flat. However, with theincrease of scanning step, for example from 4 to 8 mm, thecalculated flatness value decreases from 99.82 to 55.38%,which means there are gentle undulations on the coatedsurface. As reported in Ref 12, good deposition uniformitycan be achieved by specifying constant velocity motionwith a track gap distance of one standard deviation of thecorrected distribution. The standard deviation of Gaussiancurve used in Table 2 is 3, thus the simulation result isaccordance with the ‘‘one standard deviation’’ rule ofthumb in Ref 12. The nonlinear relationship betweencoating thickness and spray distance is shown in Fig. 8.Theoretically, the scanning steps are divided into twozones: zone A and zone B. In zone A, such as the scanningstep changes from 2 to 3 mm in this study, the homoge-neity of coated surface keeps prefect. In zone B, the cal-culated flatness of coated surface decreased obviously,which means the scanning step values can only be used ingross tolerance.

The calculated thickness value helps to approximatelyestimate the cycle number and scanning step under theexpected coating thickness.

Another experiment was carried out under the sameexperimental parameters (gun velocity = 20 mm/s) inorder to check the precision of simulated coating thick-ness, different scanning steps changing from 2 to 7 mmwere implemented in the experiment.

The measurement on sprayed sample was then per-formed in order to be compared with the simulatedthickness. Figure 9 shows the profiles of real coatingthickness (blue curves) and simulated coating thickness(pink curves).

Fig. 8 Relationship between coating thickness and step length

360—Volume 23(3) February 2014 Journal of Thermal Spray Technology

Peer

Revie

wed

In Fig. 9, the variation between pink curve and bluecurve represents the trivial error between simulatedthickness and real thickness, which is probably caused bythe measurement errors or experimental errors (such ascurrent fluctuation). The average error between simulatedthickness and real thickness at the scanning step of 2 is0.06 mm, and at the scanning step of 4 mm, the averageerror is lower than 0.09 mm and at the scanning step of 5,7 mm, the average error is lower than 0.08, 0.05 mmseparately. This confirms that the simulation method andthe simulated result can be accepted with certain toler-ance.

TST offers an approximate method to simulate 2Dcoating profile consisting of different materials to calcu-late coating thickness based on different scanning steps,spray distances, and to provide recommended experi-mental parameters. It helps to generate the robot trajec-tory and to determine accuracy coating parameters.

5. Effect of Spray Distance and ScanningStep on the Coating Profile

Spray distance and scanning step influence significantlythe coating profile in cold spray process. The different spraydistances can cause the variation of 2D coating profile. Thesimulation of coating curves in Matlab permits to mathe-matically analyze this variation. For a given material, thereexists an optimized spray distance with which one canobtain a high deposition efficiency. The best depositionefficiency can be obtained by calculating the 2D coating

surfaces under different spray distances. Scanning step isalso an important parameter which influences coatingthickness and coating profile. Two zones exist in the rangeof scanning step values: in zone A, the homogeneity ofcoating thickness is perfect; and in zone B, the flatness ofcoating surface varies obviously, which means the relativescanning step can only be used in gross tolerance.

6. Conclusions

In this work, a method for simulating coating profile isapplied; 2D coating profile is fitted as Gaussian curve inMatlab. The relationships between spray distance, scan-ning step, and coating thickness, deposition efficiency aredescribed. Based on these researches, a new toolkit inte-grated in RobotStudio� is developed to simulate the 2Dcoating profile, coating thickness, and flatness. It permitsto integrate the experimental parameters and results intoinside library in order to build the knowledge basis andassist users to determine operational parameters. Theproposed methodology is implemented in the cold spraysystem, and is also applicable to other spray systems. Therecommended method can be used to fulfill the increasingrequirements for high-accuracy thermal spray process.

References

1. L. Pawlowski, The Science and Engineering of Thermal SprayCoatings, Wiley, New York, 1995, p 414

2. A. Papyrin, Cold spray Technology, Adv. Mater. Process., 2001,159, p 49-51

Fig. 9 The difference between real coating thickness and simulated coating thickness

Journal of Thermal Spray Technology Volume 23(3) February 2014—361

Peer

Revie

wed

3. A.P. Alkimov, V.F. Kosarev, and A.N. Papyrin, A Method of ColdGas Dynamic Coating, Dokl Akad Nauk SSSR, 1990, 315, p 1062-1065

4. P. Fauchais, A. Vardelle, and B. Dussoubs, Quo Vadis ThermalSpraying, Thermal Spray 2011: New Surfaces for a New Millen-nium, C.C. Berndt, K.A. Khor, and E.F. Lugscheider, Ed., ASMInternational, Materials Park, 2001,

5. P. Fauchais, M. Fukumoto, A. Vardelle, and M. Vardelle,Knowledge Concerning Splat Formation: An Invited Review,J. Therm. Spray Technol., 2004, 13, p 337-360

6. F.-I. Trifa, G. Montavon, and C. Coddet, On the RelationshipsBetween the Geometric Processing Parameters of APS and theAL2O3-TiO2 Deposit Shapes, Surf. Coat. Technol., 2005, 195, p 54-69

7. A. Kout and H. Muller, Parameter Optimization for SprayCoating, Adv. Eng. Softw., 2009, 40, p 1078-1086

8. K.G. Shin and N.D. Mckay, A Dynamic Programming Approachto Trajectory Planning of Robotic Manipulators, IEEE Trans.Autom. Control, 1986, 31(6), p 491

9. S.H. Leigh and C.C. Berndt, Evaluation of Off-Angle ThermalSpray, Surf. Coat. Technol., 1997, 89, p 213-224

10. S.-H. Suh, I.-K. Woo, S.-K. Woo, and S.-K. Noh, Development ofAn Automatic Trajectory Planning System (ATPS) for SprayPainting Robots, Proceedings of the 1991 IEEE InternationalConference on Robotics and Automation Sacramento, California,1991

11. C.-J. Li, W.-Y. Li, and Y.-Y. Wang, Effect of Spray Angle onCoating Characteristics in Cold Spray, Advancing the Science andApplying the Technology, 2003, p 91-96

12. M.M. Fasching, F.B. Prinz, and L.E. Weiss, Planning RoboticTrajectories for Thermal Spray Shape Coating, J. Therm. SprayTechnol., 1993, 2(1), p 45

13. MSDN, Microsoft� Company, www.msdn.com14. ABB API documentation, Robotstudio� User�s Manual, ABB

Company15. J. Peter, (1998), Introduction to Expert Systems (3 edn.), Addison

Wesley, p. 2, ISBN 978-0-201-87686-416. S. Deng, H. Li, H. Liao, and C. Coddet, LERMPS, New Func-

tions of Thermal Spray Toolkit Software Developed for Off-Lineand Rapid Robot Programming, ITSC Conference

362—Volume 23(3) February 2014 Journal of Thermal Spray Technology

Peer

Revie

wed